|Publication number||US7154872 B2|
|Application number||US 10/132,454|
|Publication date||Dec 26, 2006|
|Filing date||Apr 26, 2002|
|Priority date||Apr 26, 2002|
|Also published as||US20030202488|
|Publication number||10132454, 132454, US 7154872 B2, US 7154872B2, US-B2-7154872, US7154872 B2, US7154872B2|
|Original Assignee||Lucent Technologies Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (3), Referenced by (16), Classifications (13), Legal Events (3)|
|External Links: USPTO, USPTO Assignment, Espacenet|
1. Field of the Invention
The present invention generally relates to communications systems and more particularly to a method and system for improving tracker performance in code division multiple access (CDMA) communication systems.
2. Description of Related Art
The IS-95 standard defines features of what is frequently called a second generation code division multiple access (CDMA) communication system, a type of Direct Sequence spread spectrum modulation. An IS-95 system communicates information from a base station to mobile stations over a series of traffic channels. These traffic channels transmit and receive information, spread with a traffic channel pseudonoise (PN) code, unique to each mobile station. Using precise timing and phase information derived from a pilot channel, the mobile station is able to acquire a setup channel, and eventually, the overall system time. With this system time, the mobile station is able to differentiate between base stations and synchronize its demodulation circuitry with sufficient accuracy to recover the received traffic channel message.
In a third-generation Universal Mobile Telecommunication System (UMTS), base stations are not closely synchronized. Although mobile stations attempt to lock the base station carrier frequency, the operation is imperfect such that a frequency offset is present between a mobile station and a corresponding base station. The frequency offset becomes more obvious when the mobile station is in a handoff mode (a handoff mode is where a mobile station talks to two unsynchronized base stations) or the mobile station moves at a higher speed, which causes higher Doppler frequency. Without performing some type of timing error correction, an on-time error in a base station receiver can increase with time due to this aforementioned frequency offset. An on-time error is a timing error that represents a time difference between a detected transmission path, or a finger, and the actual transmission path.
To combat and correct these on-time errors, base station receivers employ what is called a tracker. In the case where a frequency offset exists between a mobile transmitter and base station RAKE receiver, a finger (e.g., detected propagation path from mobile station transmitter to base station receiver) in the base station receiver experiences a certain adjustment in slew rate. The slew rate can be defined as a finger timing change rate in a chip per radio frame. The slew rate due to the frequency offset is such that an ideal on-time (i.e., ideal finger timing) drifts over time. Thus, a tracker should be able to track the timing drift and maintain on-time error as small as possible.
Accordingly, a tracker is needed for each finger in the RAKE receiver of the base station to track the correct timing of a specific finger. Due to a Doppler effect and due to the above-noted frequency offset between the transmitter and the receiver, the timing of a finger may drift over time. The tracker is supposed to constantly lock on a finger's timing so that most of signal energy in that finger can be utilized by the receiver.
Accordingly, given an on-time value, the tracker 100 reads two samples per chip that are output from the circular buffer 120. One sample, which is called the “early sample”, is at a timing of on-time, minus ½ chip period and the other sample (the “late sample”) is retrieved at a timing of on-time plus ½ chip period. These samples accumulate in an early accumulator 122 and in a late accumulator 124, respectively as the circular buffer 120 is filling up, wherein samples of an entire frame of data fill circular buffer 120 at a faster rate than the samples that are input into interpolator 110 from sample buffer 105.
Considering a single raised cosine pulse without noise, the early and late samples should have the same magnitude, if the on-time value is at the peak of the pulse. However, if the on-time value is earlier than the peak time, the early sample magnitude should be smaller than the late sample magnitude. Similarly, if the on-time value is later than the peak time, the early sample magnitude should be larger than the late sample magnitude.
Tracker 100 separately calculates metrics of a frame. A metric is an accumulated signal in a frame that is representative of the early or late samples. L2s shown in accumulators 122 and 124 of
A difference (d) between the early and late metrics in a frame is calculated and accumulated in a calculator 130 as a variable D. At the end of each frame, D is compared in a comparator 140 with a fixed threshold value (thresh). If the absolute value of D is larger than thresh, which means the error is significant, a first error correction unit 152 in adjuster 150 performs an error correction by adjusting the on-time error value by plus or minus 1/16 chip period (Tc), depending on the sign (+ or −) of D. A second correction unit 154 reduces the magnitude of D by the threshold value thresh. The adjustments to on-time errors are then fed back from adjuster 150 to both the sample buffer 105 and interpolator 110 for a subsequent data frame.
The IS-95 standard algorithm described above for tracker 100 has been shown to be a very good performer in the field. However, for a system operating according to the third-generation UMTS standard, this design is inadequate for two reasons. First, receivers such as RAKE receivers in a UMTS need to cover a much wider range of signal-to-noise ratios than other communication systems, due to the various data rates supported by the standard. Thus, the expected values of d and D in
Additionally, the threshold value should be adapted to handle different slot formats and compressed mode patterns, which are design considerations unique to UMTS systems, since all of these factors change the statistical properties of d and D. Therefore, in order to cover all transmission scenarios, a large number of thresholds need to be generated and stored in a lookup table. This process of generating each threshold individually and selecting a threshold on the fly requires a complicated and costly tracker design.
A second problem with the conventional IS-95 tracker is with the accumulation of d, the difference value between early and late metrics. In a UMTS standard, one transmission may include more than one slot format for different frames. If metrics calculated from different slot formats are accumulated together, the thresholds designed for a particular slot format may not work adequately, and tracker performance may therefore degrade significantly.
To overcome the aforementioned problems, the present invention provides a method and system for tracking and correcting on-time errors of fingers in a CDMA communication system. In the method, a first metric representing a first signal sample and a second metric representing a second signal sample are produced from sample signals of a data frame received in the system. A sum of the first and second metrics and a difference between the first and second metrics are calculated. The sum is used to determine at least one threshold value, and a comparator compares the difference to the calculated threshold value to determine an on-time error for a finger. Based on the determined on-time error, an adjuster adjusts timing for the on-time error, and feeds back the revised on-time value to the buffer. The revised on-time value is then applied to subsequently received data frames.
In an aspect of the invention, two threshold values are dynamically calculated, i.e., in real time. Based on a comparison of the calculated on-time error with one of the dynamic thresholds, and also accounting for the magnitude of the on-time error, a timing adjustment is performed.
The present invention will become more fully understood from the detailed description given hereinbelow and the accompanying drawings, wherein like elements are represented by like reference numerals, which are given by way of illustration only and thus are not limitative of the present invention and wherein:
The present invention is applicable, but not limited to, UMTS base station RAKE receivers. For example, the present invention may be adapted to IS-95 standard systems. The method has very low complexity without compromising system performance. Simulations illustrate that the method and system are robust for data frames having different signal-to-noise ratios, and for metrics calculated from multiple and differing slot formats. Additionally, the method employs an adaptive or dynamic threshold technique to accommodate different transmission scenarios.
These early and late symbols are accumulated so that the detector can generate metrics (Step S20). The detector generates a first metric representing the early symbols, and hence its corresponding signal samples, and a second metric representing the late symbols and hence its corresponding signal samples.
Based on the generated metrics, a sum (M) of the two metrics and a difference (D) between the two metrics are calculated (Step S30) and output from the detector. Based on the sum M, a calculator in the tracker calculates thresholds (Step S40) that are used for comparison to the difference D, which is representative of the on-time error between the two metrics.
In an embodiment, two thresholds may be calculated, a first threshold (thresh 1) for evaluating smaller on-time errors, and a second threshold (thresh 2) for evaluating larger on-time errors. If the absolute value of D is greater than thresh 2 (YES at Step S50), a larger timing adjustment is made to the on-time error (Step S55). If the absolute value of D is less than thresh 2 (No at Step S50), then the absolute value of D is compared to thresh 1 at Step S60. If the absolute value of D is greater than thresh 1 (YES at Step S60), then a smaller, refined timing adjustment, as will be discussed in detail below, is made (Step S65). In both cases, a timing error control value that represents an on-time error correction is fed back to the buffer for a subsequently received data frame (Step S70). Where the absolute value of D is less than thresh 1 (NO at Step S60), no timing adjustment is made (Step S75) for that data frame.
The tracker of the present invention includes two major parts. The first part is a timing error detector (TED), which accepts inputs of early and late symbols and calculates early and late metrics. Outputs of the TED include the difference and sum of the early and late metrics.
The second part of the tracker performs threshold comparison and timing correction, where TED outputs are compared with thresholds and sub-chip timing corrections to on-time errors are adjusted according to the comparison results. In addition, control and timing of the tracker are particularly important because of the wide varieties of operating modes in UMTS.
The TED generates early and late metrics. Early and late metrics are first calculated over a slot, and then accumulated non-coherently with other slots. Early symbols are defined as complex numbers that represent corresponding signal samples of a data frame received in the detector. Early symbols are input into the TED at a symbol rate. As previously discussed, these symbols are generated from despread DPCCH samples that are a half-chip time earlier than the “on-time” samples, which are samples obtained using the current path delay value of a finger at the receiver.
Likewise, late symbols are created from samples a half-chip later than the on-time samples. Early and late symbols are accumulated separately by coherent or non-coherent accumulation. If a series of continuous symbols in a slot are known a priori, these symbols are demodulated and then coherently accumulated. After that, results of coherent accumulation and symbols not qualified for coherent accumulation are non-coherently accumulated over a slot period.
The accumulation process is better explained in the following example. Assuming that a slot format 2 of a UMTS DPCCH channel is under consideration and Se(1), Se(2) . . . , Se(10) represent early DPCCH symbols in a slot, the metric of this slot is calculated as follows:
The first five symbols are accumulated coherently because these are pilot channel symbols and have been already demodulated to be in phase with each other. The next two symbols 6 and 7, called Transmit Power Control (TPC) symbols, are not known a priori, but they are always the same by UMTS standard. Therefore, symbols 6 and 7 can be coherently accumulated without demodulation. Because symbols 8 and 9, called Transport Format Combination Indicator (TCFI) symbols and symbol 10 (a Feed Back Information (FBI) symbol) cannot be accumulated coherently with other symbols, the final slot metric is a non-coherent accumulation of Pilot, TPC, TFCI1, TFCI2 and FBI. A metric L2 norm, shown as absolute-squared in equation (1), converts the coherent accumulation results to be ready for non-coherent metric accumulation in the TED. Metrics of different slots are also non-coherently accumulated by a direct sum.
There is an exception regarding the coherent accumulation of pilot channel symbols. The purpose of doing coherent accumulation is to reduce noise power and thus increase signal-to-noise ratio. However in the presence of non-zero Doppler frequencies and frequency offsets, demodulated symbols have intrinsic phase offsets. Thus, signal-to-noise ratio gains obtained from coherent accumulation are compromised. If the sum of Doppler frequency and frequency offset is large, the length of coherent accumulation interval should be shortened to avoid further Signal to Interference Ratio (SIR) loss. In that case, the coherent accumulation interval for pilot channel symbols should be no more than 4 symbols. In other words, if a slot has more than four (4) pilot channel symbols, the TED only coherently accumulates 4 pilot symbols and then non-coherently accumulates the rest of the pilot channel symbols. For the above example, the slot metric is computed by expression (2):
Accordingly, when it is time for the tracker to perform a timing adjustment (usually this is at the end of a frame) the TED outputs a sum (M) and a difference (D) of the accumulated early and late metrics, and all accumulators in the TED are then reset to zero for the next round of accumulation (for a next frame, for example). Based on M, two thresholds are dynamically calculated, i.e., calculated in real time. As opposed to selecting thresholds that have been calculated in advance for different conditions, and then stored in a lookup table (LUT); the present invention determines these thresholds in real time, therefore dispensing with the use or necessity of a large or complex LUT. The two threshold values may be computed using the following expression (3):
Thresh1=(M−N F)/8, and
Thresh2=3*Thresh 1. (3)
In expression (3), NF is the estimated noise power factor within an entire duration of accumulation, and M is the sum of the early and late metrics. Subtracting NF from M reduces the bias of M due to the noise factor. The remainder of this operation, i.e., the comparison of D with the thresholds, can be explained using the following pseudo code in expression (4):
Else if (abs(D)>Thresh1)
On-time−=sign(D)*1/16chip period. (4)
An on-time error adjustment of either 1/16 chip period (for smaller on-time errors) or 2/16 chip period (for larger on-time errors) is realized by setting a front-end sample buffer and interpolator of a RAKE receiver to retrieve samples with the new on-time value.
Outputs of the TED 310 include a difference D and a sum M of the early and late metrics. The sum M is used to calculate the real-time thresholds at a calculator 320. The difference D is compared to thresh2 at comparator 330 which receives the calculated thresh2 as an input. Comparator 330 also receives the calculated thresh1 as an input, to be compared with D should the on-time error be small (|D|<thresh2). Based on the results of the comparison(s), sub-chip timing correction to on-time errors is performed in adjuster 340 according to the comparison results.
Since timing control is important for the receiver in the base station, the outputs of adjuster 340, which include any corrections to the on-time error, are fed back to the sample buffer 305 and interpolator 310 for subsequently received frames. Additionally, Threshold1 and Threshold2 are dynamically (i.e., in real-time) updated for each subsequently received data frame.
One DPDCH (dedicated physical data channel) was transmitted in parallel with a DPCCH channel carrying random information, and was spread by 256 chips per symbol. The chips were then filtered through a pulse-shaping filter, which is a root raise cosine (RRCOS) filter having a 0.22 roll-off rate. This RRCOS FIR filter was six chips long and had eight samples per chip. The channel introduced additive white Gaussian noise (AWGN) with fixed power (thus a given DPCCH carrier signal to noise ratio (Ec/No) was achieved by adjusting the signal power). When Rayleigh fading was activated, the same channel fading coefficient was applied to a one-chip duration, in order to reduce computation complexity of the simulator.
At the receiver side, the same RRCOS FIR filter was implemented in one polyphase filter to convert the incoming 8 samples/chip data stream into a 16 samples/chip data stream. This is not how an actual receiver front end works, but the simulator provided similar 16 samples/chip data that is required by the tracker. All simulator runs were performed on a slot-by-slot basis.
Only one propagation path was considered in the simulation. The propagation path could be a fading path or a non-fading path, a static path or a moving path. For the following simulation results, frequency offsets were either set to zero, or in a worst case were set such that
where fo is the frequency offset; and v is the vehicle speed in km per hour. The finger slew rate is derived from the frequency offset as
Given that a UMTS radio frame is 10 ms long, the slew rate can be written as:
slew rate=fo·1.92·10−5 chip periods/frame. (7)
To reflect a given slew rate in the simulator, an accumulator was set to accumulate the timing drift every one eighth (⅛) of a chip period. When the accumulator reached 1/16 chip period at the end of a slot, a timing offset of one sixteenth chip period was introduced to the next slot. The offset could be either plus or minus depending on the sign of the frequency offset.
The following simulations in
In the simulation, if the Doppler frequency was higher than 250 Hz, the propagation path was considered as a fast fading path. Otherwise it was considered a slow fading path. All of the simulations started with a zero on-time error, so the initial transient effect was not included in the results.
Slot format 2 of the UMTS standard was considered next. While slot format 1 of the UMTS standard had eight pilot symbols, slot format 2 of the UMTS standard has only five pilot symbols, so some performance degradation due to less noise suppression was expected.
The invention being thus described, it will be obvious that the same may be varied in many ways. For example, the functional blocks in
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US6269075 *||Jan 26, 1998||Jul 31, 2001||Nokia Mobile Phones Limited||Finger assignment in a CDMA rake receiver|
|US6735242 *||Aug 30, 1999||May 11, 2004||Nokia Corporation||Time tracking loop for pilot aided direct sequence spread spectrum systems|
|US20040165655 *||Feb 26, 2004||Aug 26, 2004||Rick Roland Reinhard||Time tracking in a non-negligible multipath spacing environment|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7272167 *||Aug 3, 2006||Sep 18, 2007||Neoreach, Inc.||PN code chip time tracking with smart antenna|
|US7382744 *||Jan 27, 2005||Jun 3, 2008||Qualcomm Incorporated||Systems and methods for optimizing the allocation of resources to serve different types of data flows in a wireless communication system|
|US7515876 *||May 3, 2005||Apr 7, 2009||Agere Systems Inc.||Rake receiver with time-shared fingers|
|US7756481||Mar 25, 2009||Jul 13, 2010||Agere Systems Inc.||Rake receiver with time-shared fingers|
|US7869486 *||Aug 18, 2005||Jan 11, 2011||Samsung Electronics Co., Ltd.||Tracking apparatus and method for a mobile communication system|
|US8170087 *||Sep 28, 2007||May 1, 2012||Texas Instruments Incorporated||Correlation coprocessor|
|US8594072 *||Aug 4, 2010||Nov 26, 2013||Qualcomm Incorporated||User equipment based method to improve synchronization shift command convergence in TD-SCDMA uplink synchronization|
|US8619836 *||Jan 30, 2012||Dec 31, 2013||Texas Instruments Incorporated||Correlation coprocessor|
|US20050078741 *||Aug 9, 2004||Apr 14, 2005||Hun-Geun Song||Apparatus and method for detecting a timing error in a mobile communication system|
|US20060039452 *||Aug 18, 2005||Feb 23, 2006||Samsung Electronics Co., Ltd.||Tracking apparatus and method for a mobile communication system|
|US20060164990 *||Jan 27, 2005||Jul 27, 2006||Naga Bhushan||Systems and methods for optimizing the allocation of resources to serve different types of data flows in a wireless communication system|
|US20060251155 *||May 3, 2005||Nov 9, 2006||Gongyu Zhou||Rake receiver with time-shared fingers|
|US20060268964 *||Aug 3, 2006||Nov 30, 2006||Song Yoo S||PN code chip time tracking with smart antenna|
|US20080279261 *||Sep 28, 2007||Nov 13, 2008||Texas Instruments Incorporated||Correlation coprocessor|
|US20090180523 *||Mar 25, 2009||Jul 16, 2009||Agere Systems Inc.||Rake receiver with time-shared fingers|
|US20110243121 *||Aug 4, 2010||Oct 6, 2011||Tom Chin||User Equipment Based Method to Improve Synchronization Shift Command Convergence in TD-SCDMA Uplink Synchronization|
|U.S. Classification||370/335, 375/E01.016, 375/E01.032, 375/150, 370/342|
|International Classification||H04B7/216, H04B1/707, H04L7/02|
|Cooperative Classification||H04L7/0054, H04B1/7085, H04B1/7117|
|European Classification||H04B1/7085, H04L7/00D|
|Apr 26, 2002||AS||Assignment|
Owner name: LUCENT TECHNOLOGIES INC., NEW JERSEY
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:HSUAN, YI;REEL/FRAME:012840/0303
Effective date: 20020424
|Jun 17, 2010||FPAY||Fee payment|
Year of fee payment: 4
|Jun 19, 2014||FPAY||Fee payment|
Year of fee payment: 8